Keith Watanabe * NET 2.0

Every year you'll see a discussion from the OSS community regarding whether or not Linux as a Desktop is ready for prime time.  Certainly, rumblings for 2008 will become apparent with each new release from distributions like Fedora, Ubuntu, and PCLinuxOS.  While tons of improvements have been made to the Linux OS and the various tools involved, my primary complaint is that it's still too developer oriented. 

Let's face it.  Most of us who use Linux as a tool are developers.  Good development environments often implement Linux on the desktop and virtually most of the tech groups employ Linux as their main OS of choice (Mac OS comes in a close second).  It's funny to think then that those same environments where the techies understand what a great OS Linux is fail to convince management to standardize it across the company.  So strike one.  Not business friendly apparently.

From an at-home viewpoint, things aren't much better.  It reminds me of Apple's commercial bashing Windows 95 or 98 back when Steve Jobs/Apple were making their comeback.  It was funny actually because you had this regular guy at home going, "What is this io.sys file?"  Hey people, why give something such a mysterious name like "grub?"  You think these names make it any easier for someone coming from a "it just works" background to a "I know how to fix my own plumbing!" background?  Transparency, transparency, transparency.  Developers need flexibility but home users need transparency.

Next the number of tools are enormous.  It's great to have choices and avoid M$-like monopolies.  But in all honesty, most people just need a few tools to get their work done.  It seems whenever someone gets pissed off that a certain tool doesn't have a feature they want, they scurry into their dank hovel and build another one just for the sake of the feature.  That doesn't seem that efficient.  And to try and figure out which tools to employ is pretty brutal because the number of tools in something like synaptec is downright mind boggling!  How can one attempt to figure out what the best media player is?  Totem?  Mplayer?  XMMS?  Xine?  In Windows, you just use Media Player and it handles most of your tasks.  While I'm somewhat familiar with a number of the programs available from synaptec, imagine how a newbie might be intimidated by the availability of programs there.

Next where are all the games?  Yes, there are games for Linux but they suck!  Well, I don't think the PC game market is thriving that much anyway since you have the DS, PS and XBoxes roaming out there.  But believe it or not, there are some people like myself who will actually PAY for games developed on a Linux box just to avoid having to start up my Windows piece of garbage.

Just simple compatibility problems from browsers.  This seems like a small one, but it's really frustrating going to a site that's IE only or that renders funny on Linux.  The thing is that people just don't test for compatibility on Linux or really don't care to when it comes to websites.  Fortunately, many sites render fine but as we move towards richer web applications, I'm afraid that this issue will get worse.

Better hardware support.  It's come a LONG, LONG WAY since the days where even a modem would be difficult to get working.  But they really need to resolve compatibility problems.  ATI video cards?  This isn't as much fault with the Linux community as it is with the hardware manufacturers.  But it's a two way street since the hardware manufacturers probably still see Linux as a server side component with spots of life on the desktop side.  Therefore to them there's not really a huge business case (at the moment) to invest more money into testing and the extra coders.  So you could say that this is a chicken-or-the-egg type of problem.

Even if all the most useful applications move to the web, I seriously doubt that Linux will become a force on the desktop, at least in the vision that most people probably want.  Maybe someone might invent a "dumb client" where you just have a browser with flash.  But by that point, Linux won't have any real use as a desktop.  Think about that point long and hard.

I'm certain that most people in the Linux community recognize these points as the chief ones in helping Linux penetrate the desktop market (except the one about it being too developer oriented).  However, it never can be understated that these have to be addressed over and over again.

I think one inroad that Linux can make against M$ and Mac OS is determining what a person needs before they need it.  This is sort of B. Gates' philosophy for M$, except his aim was generally for businesses.  Whenever you get a newly installed Linux system (or as you're installing Linux), you either get some minimal working environment where you're forced to grab a ton of packages, taking a few more hours, or you're force to grab a ton of packages from a menu and that most of the packages aren't really the ones you're looking for (because of legal reasons).  I see this methodology as being ass backwards.

I kinda like the minimal install idea because you really don't want a lot of garbage floating around your system.  But you shouldn't have to choose between a million packages, especially if you don't know what you want to do.  One major problem I see in Linux and its design is in the assumptions made about the user.  Most of the time, there are either no assumptions being made, or that the user is some silly Windows convert, or that the person is a guru.  You can't market something without assumptions so this item must be crossed off the list.  Probably most of the Linux system is designed for the "average" (read: guru) user.  And when Linux tries to do Windows (or should I say a bad Mac impersonation) it's just plain bad. 

There should be (simply stated) two modes of operation: idiot and guru.  In my world, I'd even assign points/levels as the person makes progress.  But you have to design for idiots because the world is composed of them (including myself).  That means, you have to go back to fundamental questions in designing your interfaces, since this is what people have to work with.  The most fundamental question is, what do you want your users want to do with this thing?

Before trying to re-create Windows, Mac OS or anything, answer that question first.  Why should people go to a website and download a specific distribution for servers, desktops, or alternatives?  What difference does it really make?  The download should NOT determine the user's preference (because people can tend to become confused on this purpose).

I think there should be just two distribution types: guru and idiot.  The guru version has no wizards and is just what it is at the moment.  The idiot version is the one that the Linux community needs to work on.  Ever see Winzip and how they've done their interface?  It's great.  You have a wizard (idiot) or classic (guru).  Or have you used TurboTax?  Again, guru (fill out everything by hand) or idiot (tons of questions that you need to answer).

I know Fedora tried something like this, but even I'm skeptical.  To me, there were too many questions and choices.  Worse yet, some were plain obvious.  I mean, why not install OpenOffice?  You don't see the ipod or iPhone asking questions on what applications need to available when you receive one.

I think the killer application for Linux would be to automatically figure out and get something to run, load, view, etc. a file based on the extension type and headers.  Imagine encountering an mp3 for the first time.  Certainly, most distributions for legal reasons will not ship an out-of-the-box codec.  But why should a home user be forced to go through hoops just to listen to music?  If a home user was browsing his file system via Konquer and double clicked on the file, if the application did not already exist on the system, a little popup should appear saying, "Installing XXX...." making some specific assumptions about what the user wants based on the most popular viewer or whatever for that application.  In the case of the mp3 file, a legal disclaimer could popup and get the user to check it off before going out and grabbing the correct codec.

Next, forget the whole menu system.  There should be a question based system based on the most commonly used applications on the system.  You could have it say things like, "I want to listen to music," or "I want to watch a movie," or "I want to read my email".  As people use more applications, these questions might disappear and form shortcuts, learning about the user, rather than the user being forced to learn about the system.

This is a big thing because really menu systems in general have sucked for the longest time.  In the case of linux, even with synaptec, you still can get a boatload of confusion.  For instance, after installing something like multi-gnome-terminal, where the hell in the menu is it supposed to be located?  Should the application be readily available after I installed it?  It's again not transparent.

Overall, I seriously doubt that Linux will adjust itself based on my recommendations.  They are very much valid and should be considered heavily.  Instead of imitating, the Linux desktop needs to innovate.  People who work on things like installation or KDE need to focus more on how the average person can arrive to the answers of their problem, rather than just writing a million applications that most people won't ever use.  Copy the core philosophy that Apple has, not just the appearance of it.